GFAP promoter directs astrocyte-specific expression in transgenic mice

Brenner et al. (1994). J Neurosci DOI: 10.1523/JNEUROSCI.14-03-01030.1994 Citations: 500

Astrocyte-specific expression in cultured cells did not guarantee specificity in an animal; 2.2 kb of human GFAP 5′-flanking sequence confined a lacZ reporter to CNS astrocytes in transgenic mice and reproduced the GFAP upregulation after brain injury.

Parts used

Key findings

  • A 2.2 kb fragment of the human GFAP gene spanning base pairs −2163 to +47 relative to the transcription start site contains sufficient regulatory information to direct astrocyte-specific expression in transgenic mice, not merely in cultured cells. The fragment used carries the native protein-initiating ATG at +15 mutated to TTG so the reporter is translated instead.
  • Expression from the human GFAP promoter in transgenic mice is confined almost entirely to brain. In a quantitative β-galactosidase solution assay across brain, kidney, liver, spleen, heart and lung of adult animals, brain is the only tissue with activity significantly above nontransgenic controls, reaching up to 80-fold higher, and histochemical staining is unsuitable for surveying other tissues because of high background in some organs, particularly kidney.
  • The GFAP promoter drives expression throughout the CNS gray and white matter, including the cerebellum, brainstem and cerebral hemispheres, with distinct labelling of Bergmann glial cell bodies and their radial processes in the cerebellar molecular layer. This pattern is reproduced in all six expressing transgenic lines, differing between lines only in staining intensity.
  • Ultrastructural analysis establishes that the human GFAP promoter is astrocyte-specific at the level of individual cells rather than merely enriched. In optic nerves from three transgenic lines, with over 50 astrocytes and oligodendrocytes identified morphologically and the electron micrographs coded and scored by an independent expert, most but not all astrocytes were labelled by the β-galactosidase substrate Bluo-gal while no other cell type was labelled at all.
  • Apparent GFAP-negative but reporter-positive cells in thick stained sections reflect limited antibody penetration rather than genuine non-astrocytic expression: double labelling shows many cells with X-gal-positive nuclei surrounded by GFAP-immunoreactive processes, but reporter-positive GFAP-negative cells concentrate toward the centres of slices, and ultrastructural analysis of the same tissue finds no labelled non-astrocytes.
  • The 2.2 kb human GFAP promoter fragment retains the regulatory elements responsible for injury-induced upregulation, reproducing reactive gliosis at the transcriptional level. A stab wound to the cerebral cortex produces a marked increase in reporter staining in the immediate vicinity of the needle track within 48 hours, in all three transgenic lines tested.
  • Reporter expression from the human GFAP promoter follows the developmental profile of the endogenous mouse GFAP gene postnatally but begins earlier: transgene activity is already detectable in the fetus at embryonic day 13.5, whereas the earliest reported detection of GFAP mRNA or protein is about embryonic day 16.5 in both mouse and rat.
  • The earliest sites of human GFAP promoter activity in the embryo are neuroepithelial progenitor zones rather than differentiated astrocytes. Expression appears between embryonic days 12.5 and 13.5 in each of five lines examined, localized to the ventricular zone of the developing telencephalon and hippocampus, the hindbrain, and cells lying dorsally around the central canal of the spinal cord extending into the developing dorsal median sulcus, together with a dorsal midline streak. By embryonic day 17.5 staining is found throughout the brain, most intensely in the developing cerebellum and at ventricular surfaces, becoming uniform across gray and white matter later.
  • Not every transgenic founder line expresses, and insertion site matters: of ten transgenic founders identified by PCR, eight produced breeding lines and six of those eight expressed lacZ in the brain, with two lines staining less strongly than the others.
  • An X-linked GFAP-lacZ transgene undergoes normal random X-inactivation, giving hemizygous females half the brain β-galactosidase activity of age-matched males — yet their brains stain uniformly, with no mosaic patches detectable at a resolution able to resolve patches as small as 25 cells. Astrocytes therefore intermingle extensively after they are generated rather than remaining as clonal patches.

Abstract

Glial fibrillary acidic protein (GFAP) is an intermediate-filament protein expressed abundantly and almost exclusively in astrocytes of the CNS. We are studying transcriptional regulation of the GFAP gene to gain insight into astrocyte function and also to develop an astrocyte-specific expression system for manipulating brain physiology. In this work, we have produced transgenic mice carrying the bacterial lacZ reporter gene linked to a 2.2 kilobase 5'-flanking sequence derived from the human GFAP gene that previously was shown to direct astrocyte-specific transcription in cultured cells. We report that this promoter directs expression to astrocytes in the CNS. In addition, the upregulation of GFAP gene activity that follows injury to the brain was mimicked by the transgene. One of the transgenes was found to be X-linked and appeared to undergo the usual random inactivation that achieves gene dosage compensation in females. The brains of hemizygous females stained uniformly rather than displaying mosaic patches, indicating that astrocytes intermingle following their formation. The specific expression of the GFAP-lacZ transgene means that it is now possible to target expression of other heterologous genes to astrocytes in vivo, and to study the mechanisms for reactive gliosis at the DNA level.